Abstract Small molecules from microbes known as `natural products' have been an important source of clinically- approved drugs and cell biological probes, and developing new strategies for their discovery will expand the stream of natural products entering compound libraries at the NIH and elsewhere. Using a bioinformatic algorithm we recently developed to identify natural product-encoding gene clusters in microbial genome sequences, we have found numerous gene clusters in bacteria from a surprising location: the human microbiome. Herein, we propose to use a combination of chemistry, genetics, and bioinformatics to isolate, purify, and solve the structures of three classes of molecules from human gut and oral bacteria. This project will lay the groundwork for a larger effort to identify and characterize natural products from the human microbiome. There are three reasons why identifying natural products from the human microbiome should be a priority: 1) Underexplored ecological niches and bacterial phyla usually yield new natural products. 2) Natural products from the human microbiome are likely to have important biological activities, and are unlikely to be toxic to humans. 3) Computational searches for new natural products have become increasingly fruitful, and the Human Microbiome Project will provide a wealth of sequence data for human-associated bacteria. Not only will our efforts uncover natural products with unusually important biological activities, they will also validate the human microbiome as a new source of natural products and enable future studies of the molecular basis of interactions between humans and their microbiome. While this proposal is focused on identifying three classes of molecules from gut and oral bacteria, our long-term goal is to lay the groundwork for systematizing a computational search for new natural products from the human microbiome. Our efforts are organized into three Specific Aims: Aim 1: Identify and characterize the products of a family of NRPS gene clusters from gut bacteria. We have identified a family of >50 NRPS gene clusters that stands out for two reasons: 1) One or more of these gene clusters is present at a high level in 90% of healthy human subjects, making this one of the most widely distributed gene cluster families in the human population. 2) This family is only found in gut-dwelling bacterial species, raising the possibility that it plays an important biological role in the context of gut colonization. Aim 2: Identify and characterize the products of a family of NRPS and NRPS/PKS gene clusters from oral strains of Streptococcus. By analyzing the genome sequences of 285 oral isolates of Streptococcus, we have found that the major genetic differences among strains of these species are 14 biosynthetic gene clusters that occupy up to 8% of the genome. Metagenomic and metatranscriptomic analyses suggest that of these gene clusters, eight are large (up to 12- module) NRPS, PKS, and NRPS/PKS hybrids; all are widely distributed in the human population, and most are actively transcribed on teeth. Aim 3: Identify and characterize the products of a family of oligosaccharide gene clusters from Bacteroides. The largest family of biosynthetic gene clusters in the human microbiome, which are responsible for producing Bacteroides capsular polysaccharides, is underexplored. These molecules are present at high abundance in the gut and are known to have specific ligand-receptor interactions that modulate the host immune response.